Lipopolysaccharide (LPS) release from invading bacteria has been closely associated with the pathogenesis of the inflammatory tissue injury occurring during gram-negative sepsis in humans. Agents designed to inhibit endotoxin have been proposed as adjunctive therapy for sepsis. E5564, a lipid A analogue, is one such agent which has been shown to competitively inhibit LPS stimulated cytokine release from macrophages. However, although LPS signaling may stimulate inflammatory mediators harmful to the host, this response may also have an adaptive protective function. In fact, other agents (e.g. antiendotoxin antibodies) designed to inhibit LPS which were beneficial in animal models of sepsis, have not shown beneficial effects in large clinical sepsis trials. It is possible that in animal models that have frequently employed intravascular (IV) bacterial challenges, the effects of LPS on host defense and inflammatory injury are different than during the extravascular (EV) infection primarily observed in patients. For instance, the intravascular activation of leukocytes by endotoxin may have little protective effect during IV bacterial challenge but may be important for their recruitment to an EV nidus of infection. This study therefore compared the effects of E5564 with similarly lethal IV and EV infection. Rats received E5564 or placebo after IV or EV [intrabronchial or intraperitoneal] E. coli challenges. E5564 decreased the relative risk of death with IV E. coli and increased it with EV infection in patterns that were significantly different. Compared to controls, in both IV and EV E. coli, E5564 increased circulating total leukocytes and neutrophils at 4 and 24 h combined but decreased lung lavage neutrophils at 4 h while increasing them at 24 h. Thus, the ability of E5564 to impair tissue leukocyte recruitment may explain the lack of benefit or potential harm associated with the agent in this model of EV infection. Conversely, in IV infections, the same affect on leukocyte trafficking may limit non-specific organ injury and thereby improve survival. Site of infection may have an important impact on agents designed to alter LPS levels in sepsis. Clinical experience with antiinflammatory agents in patients with sepsis has been disappointing to date. We have found that several factors such as the site, type and severity of infection have important influences on many of these agents. Developing new agents which are impacted minimally by such factors, will increase the usefulness of this therapeutic approach. The role of endotoxin in the injury of sepsis is unclear. It is likely that under many circumsatnces it produces harmful effects. Targeting endotoxin rather than host mediators may be a more generally useful goal in sepsis. Antibodies against endotoxin have had little effect to date in sepsis because the toxic lipid portion of the molecule may be difficult to target. An alternative approach to neutralizing endotoxin uses analogue molecules which competitively inhibit cell signaling by endotoxin. E5564 is one such competitive inhibitor.